Theoretical and computational studies of microscopic water channels

Water channels are ubiquitous in all life forms. A notable example is aquaporins (AQPs), a family of proteins which mainly function as passive water channels in cell membranes. The availability of the crystal structures of several AQPs in recent years allowed us to study them in atomic details. We performed molecular dynamics (MD) simulations on AQPs in lipid bilayers to study water permeation through these channels. We also simulated water conduction in carbon nanotubes, serving as simplified models for biological channels. We developed theories and methodologies aimed to reproduce and predict important experimental quantities of water channels from simulations. We showed that the diffusion permeability (pd), which is measured by tracer diffusion in experiments, can be calculated from equilibrium MD simulations. In order to calculate the osmotic permeability (p f), which is experimentally measured in the presence of a solute concentration difference, we developed a method to induce a hydrostatic pressure difference across the membrane under periodic boundary conditions. We calculated the osmotic permeability for aquaporin-1 using this method, which agrees with experiments. Using a continuous-time random-walk model, we showed that for single-file water channels, the ratio of p f to pd is roughly equal to the number of water molecules in the channel. Proton transfer through single water file was studied theoretically using network thermodynamics. Finally, we proposed a new model for general water channels, which gives a quantitative relationship between water permeations under equilibrium and non-equilibrium conditions, and therefore allows one to calculate pf from equilibrium MD simulations.